How to see instances of incorrect images, labels and predictions in test data Pytorch - Fashion MNIST?

vision
1

Hi.
In the notebook below I’ve created a Deep Learning Model in Pytorch and trained and tested it.
I want to save the instances of the test image data and the test image labels which and predicted test labels which came incorrect running it on the test set.
I have been trying to implement it using the for loops near the end of the Jupyter Notebook using :

correct = 0
total = 0

incorr_X = []
incorr_y = []
incorr_argmax = []

with torch.no_grad():
    for data in testset:
        X, y = data
        output = net(X.view(-1,784))
        #print(output)
        for idx, i in enumerate(output):
            #print(torch.argmax(i), y[idx])
            if torch.argmax(i) == y[idx]:
                correct += 1
            else:
                incorr_X.append(X)
                incorr_y.append(y)
                incorr_argmax.append(torch.argmax(i))
            total += 1

print("Accuracy: ", round(correct/total, 3))
Is this the correct approach to get the incorrectly predicted image, label and predicted label?

How do I access the corresponding images, labels and predicted labels?

I have written some implementations in the Notebook, but they seem to not be working.

Any help will be greatly appreciated.

Thank you!!!

Link to Notebook:

2

Since you are iterating the output using i, you should append i in incorr_X and y[idx] in incorr_y.
Let me know, if this isn’t working and what kind of error you are seeing.